From this month's Atlantic, a few pages after the Petraeus Doctrine article:

As in Vietnam, the U.S. has never lost a tactical engagement in Afghanistan, and this tactical success is still often conflated with strategic progress. Yet the Taliban insurgency grows more intense and gains more popular traction each year. More and more, the American effort in Afghanistan resembles the Vietnam War—with its emphasis on body counts and air strikes, its cross-border sanctuaries, and its daily tactical victories that never affected the slow and eventually decisive erosion of rural support for the counterinsurgency.

As the Russian ambassador to Afghanistan, Zamir Kabulov, noted in a blunt interview with the BBC in May, the current military engagement is also beginning to look like the Soviets’ decade-long Afghan adventure, which ended ignominiously in 1989. That intervention, like the current one, was based on a strategy of administering and securing Afghanistan from urban centers such as Kabul and the provincial capitals. The Soviets held all the provincial capitals, just as we do, and sought to exert influence from there. The mujahideen stoked insurgency in the rural areas of the Pashtun south and east, just as the Taliban do now.

The U.S. engagement in Afghanistan is foundering because of the endemic failure to engage and protect rural villages, and to immunize them against insurgency.
....
To reverse its fortunes in Afghanistan, the U.S. needs to fundamentally reconfigure its operations, creating small development and security teams posted at new compounds in every district in the south and east of the country. This approach would not necessarily require adding troops, although that would help—200 district-based teams of 100 people each would require 20,000 personnel, one-third of the 60,000 foreign troops currently in the country.

Each new compound would become home to roughly 60 to 70 NATO security personnel, 30 to 40 support staff to manage logistics and supervise local development efforts, and an additional 30 to 40 Afghan National Army soldiers. The troops would provide a steady security presence, strengthen the position of tribal elders, and bolster the district police. Today, Afghan police often run away from the superior firepower of attacking Taliban forces. It’s hard to fault them—more than 900 police were killed in such attacks last year alone. But with better daily training and help only minutes away, local police would be far more likely to put up a good fight, and win. Indirectly, the daily presence of embedded police trainers would also prevent much of the police corruption that fuels resentment against the government. And regular contact at the district and village levels would greatly improve the collection and analysis of intelligence.
http://www.theatlantic.com/doc/200810/afghan

The seem to be arguing for a CAP program in Afghanistan. What was Westmoreland's main argument against CAPs, the fear that they would be overrun by superior VC/NVA forces (even though I don't think this ever happened)?